africa-ports-sierra-leone
收藏资源简介:
该数据集名为塞拉利昂:每日港口活动数据与货物运输估算,由PortWatch发布,来源于HDX平台,并由Electric Sheep Africa整理为适合机器学习的格式。数据集记录了塞拉利昂港口的每日到港次数、进港和出港货物量(以公吨计)的估算数据,每条记录代表国家层面的汇总数据。数据集包含2,650条记录,31个字段(24个数值型,6个类别型),已划分为2,120条训练数据和530条测试数据。主要字段包括日期信息(年、月、日)、港口标识与名称、国家代码、各类船只到港次数(如集装箱船、散货船等)、进出口货物量估算等。数据集适用于表格分类和回归任务,特别适合非洲贸易、港口活动分析等人道主义和发展领域的研究。数据覆盖时间为2019年至2026年,地理范围限定为塞拉利昂(SLE)。需要注意的是,数据来源于PortWatch,未经ESA独立验证,可能存在原始收集中的报告误差或定义不一致问题。
This dataset is named Sierra Leone: Daily Port Activity Data and Cargo Transport Estimates. Released by PortWatch, it is sourced from the HDX platform and formatted into machine-learning-ready structures by Electric Sheep Africa. This dataset records estimated daily port call counts, import and export cargo volumes (in metric tons) at Sierra Leonean ports, with each entry representing national-level aggregated data. The dataset contains 2,650 total records across 31 fields: 24 numerical fields and 6 categorical fields, and has been split into 2,120 training samples and 530 test samples. Key fields include date information (year, month, day), port identifiers and names, country codes, port call counts for various vessel types (e.g., container ships, bulk carriers, etc.), and estimated import/export cargo volumes, among others. This dataset is suitable for tabular classification and regression tasks, and is particularly well-suited for research in humanitarian and development fields such as African trade and port activity analysis. The dataset covers the period from 2019 to 2026, with a geographic scope limited to Sierra Leone (SLE). Notably, this data is sourced from PortWatch and has not been independently verified by ESA; report errors or inconsistent definitions may exist during the original data collection process.
数据集概况
- 名称: Sierra Leone: Daily Port Activity Data and Shipment Estimates
- 发布方: PortWatch
- 来源: HDX
- 许可协议:
hdx-other - 更新日期: 2026-04-08
- 地理范围: 塞拉利昂 (SLE)
- 语言: 英语
- 领域: 人道主义与发展数据
- 观测单位: 国家级汇总数据
- 总行数: 2,650
- 列数: 31(24个数值型,6个类别型,0个日期时间型)
数据划分
| 划分 | 样本数 |
|---|---|
| 训练集 | 2,120 |
| 测试集 | 530 |
任务类型
- 表格分类 (tabular-classification)
- 表格回归 (tabular-regression)
核心内容
数据集包含塞拉利昂港口的每日船舶到港次数、估计的进口货物量和出口货物量(以公吨计),每条记录代表国家层面的汇总数据。
主要变量
- 地理变量:
year(2019-2026),day(1-31),country(Sierra Leone),iso3(SLE) - 时间变量:
date,month(1-12) - 标识/元数据:
portid(port360),portname(Freetown),esa_source(HDX),esa_processed(2026-04-28) - 船舶到港次数:
portcalls_container(0-2),portcalls_dry_bulk(0-3),portcalls_general_cargo(0-6),portcalls_roro(0-1),portcalls_tanker(0-3),portcalls_cargo(0-8),portcalls(0-8) - 进口量 (公吨):
import_container(0-20875),import_dry_bulk(0-83287),import_general_cargo(0-17442),import_roro(0-3850),import_tanker(0-43895),import_cargo(0-83287),import(0-83287) - 出口量 (公吨):
export_container(0-14778),export_dry_bulk(0-187960),export_general_cargo(0-13153),export_roro,export_tanker,export_cargo,export
数据处理流程
原始数据通过 HDX 的 CKAN API 下载并转换为 Parquet 格式。列名统一为小写蛇形命名法,缺失值标记(如 N/A, null, none 等)统一替换为 NaN。部分列基于解析成功率(>85%)从字符串转为数值或日期时间类型。数据集以固定随机种子 (42) 按 80/20 比例随机划分为训练集和测试集,保存为 Snappy 压缩的 Parquet 文件。
局限性
- 数据源自 PortWatch,未经 Electric Sheep Africa 独立验证。
- 自动化清洗无法纠正原始数据中的错误报告、定义不一致或采样偏差。
- 建议参考原始 HDX 数据集页面获取发布方的方法说明和注意事项。
引用格式
bibtex @dataset{hdx_africa_ports_sierra_leone, title = {Sierra Leone: Daily Port Activity Data and Shipment Estimates}, author = {PortWatch}, year = {2026}, url = {https://data.humdata.org/dataset/sierra-leone-daily-port-activity-data-and-shipment-estimates}, note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)} }




